 |
|
| In My Opinion.......
ATM - Has Its Time Come and Gone? Every year when you visit a trade show you are regaled with press releases and neon signs proclaiming that this is "the year" of some technology or other. Usually, such proclamations are equalled by those claiming "the death" of another (usually rival!) technology. Unfortunately, the computer industry seems to have adopted the canine model of birthday calculation when it comes to product releases, in that one year actually equals several! For as long as I can remember, therefore, we have been faced with "The Year Of ATM". Flying in the face of the analysts and pundits � it is too expensive, it is too complicated, it will never reach the desktop, there are no standards � ATM resurfaced time after time to insist that, yes, this indeed was its "year". Surprising, therefore, to see so little evidence of it at this year�s Networld+Interop show in Las Vegas. Sure there are products to be found from the likes of Fore, Cisco, Lucent, Madge, Nortel, 3Com and IBM (amongst others), but you don�t see such a song and dance made about the technology. Listen to the analysts and you may be forgiven for thinking that the death knell has already sounded for ATM. Some have claimed that it will soon be considered legacy technology, killed by its own complexity and high prices. What is driving this accelerated demise? Amazingly enough, it is Ethernet. For years, we had a maximum of 10 Megabits to play with, and for a long time that did seem enough. The popularity of the PC coupled with huge price performance breakthroughs soon saw our networks overloaded, however, as Ethernet failed to keep pace with what was happening in the LAN world. ATM was suddenly touted as the panacea to all our problems. It would give us huge amounts of bandwidth to the desktop � from 25 Mbit/s to 2.488 Gbit/s (though the realistic financial maximum for desktop use is 155 Mbit/s) � allow us to switch all forms of traffic across our networks � voice, video and data � whilst providing something that Ethernet could never match � Quality of Service (QoS). But products were a long time appearing, and when they did they were marred by incompatibility problems. Complexities in integrating legacy networks also put many off. And all of a sudden, there was a rash of new Ethernet technologies. After a number of years of nothing but shared 10Mbit/s, we were suddenly treated to switched 10Mbit, 100Mbit and, of course, Gigabit Ethernet. Practically overnight we are seeing bandwidth to equal that of ATM over a technology we have grown to know and love. Simpler implementation and lowers costs have also endeared Gigabit Ethernet to the network administrator. But that�s not all. What else has happened to drive the Ethernet revolution (or should that be resurrection)? The Internet and all its baggage, of course, in the shape of good old TCP/IP. The whole world has gone Internet crazy and this means we need TCP/IP. Unfortunately ATM will never be any good at TCP/IP, it just wasn�t designed for it. Ethernet was, however. What about the multimedia promise of ATM? Well, that seems to have been pushed to the back by yet another technology that was very much in evidence at this year�s N+I � Voice Over IP (VOIP). Convergence seems to be on the top of everyone�s agenda this year, as the premise of free phone calls over the Internet has matured into a whole host of new applications that promise revenue generation and operational cost reductions. The "new" convergence is defined as the ability of service providers, including cable operators, to deliver voice, data and video services over a single, packet-based infrastructure. A number of vendors visible at N+I � Ascend and Nortel for instance, amongst others � are working with a wide range of service providers to create the "next generation public network", and we should finally see the new generation services appearing in the next 12 to 24 months. The one area where ATM continues to score, for the time being at least, is in QoS. New versions of IP promise improved QoS, but this will always be incredibly difficult to enforce throughout a large, distributed network. In the meantime, ATM continues to do well in the campus backbone environment with 700,000 ports shipped in 1998 and over 2 million projected by 2001 according to the Dell�Oro Group. If there is one thing that most of us have learned in this industry, it is that things usually take even longer to die than they did to appear in the first place. It remains to be seen if we will be left to consider how to make best use of our outdated, legacy ATM systems as we trek round N+I 2002.
|
|
Send mail to webmaster
with questions or�
|
